Agent skill

Generate Video

by ArcReel in ArcReel/ArcReel

为分镜或自包含视频单元生成视频。当用户要求生成或重做视频时使用;支持整集、单项与批量自选. An agent skill from ArcReel/ArcReel.

AGPL-3.0Auto-check: warningsMedia & Creative

Install Generate Video

The automated check flagged lines worth reading first. See the safety section below.

skills CLI
$ npx skills add ArcReel/ArcReel --skill generate-video -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install ArcReel/ArcReel generate-video --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/ArcReel/ArcReel.git skills-src && mkdir -p .claude/skills && cp -r skills-src/agent_runtime_profile/.claude/skills/generate-video .claude/skills/generate-video && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
generate-video
GitHub stars
5.4k
Token cost
~1.3k tokens
SKILL.md length
301 words
Files
2 (incl. references)
Skills in repo
18
Repo updated
First seen
Licence
AGPL-3.0

At a glance

为分镜或自包含视频单元生成视频。当用户要求生成或重做视频时使用;支持整集、单项与批量自选. An agent skill from ArcReel/ArcReel.

  • Works in 5 steps: 加载项目和剧本,确认骨架与生成模式一致。 → 在分镜图生视频确认分镜图可用;在参考生视频确认视频单元正文非空、编排时长合法。 → 调用 MCP 工具入队,处理准入拒绝与档位确认。 → …
  • Tasks that involve AI video generation
  • SKILL.md covers 路由, 工具调用, 工作流程 and Prompt 构建, plus 1 more section
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Generate Video is an agent skill from ArcReel/ArcReel. 为分镜或自包含视频单元生成视频。当用户要求生成或重做视频时使用;支持整集、单项与批量自选。

Its SKILL.md is about 1.3k tokens, which your agent loads only when the skill is triggered. The skill folder holds 2 other files, including reference files (for example `references/veo_prompts.md`).

It sits in Media & Creative, covering AI video generation. It works with Model Context Protocol. The repository describes itself as: AI Agent 驱动的开源可自部署视频工作台:将小说与剧本转为角色、场景、道具资产、分镜、视频和剪映草稿,支持跨镜头一致性、多供应商与费用追踪 | Self-hosted AI video workspace for stories, storyboards and short-form video production. The licence is AGPL-3.0.

When your agent uses it

  • Tasks that involve AI video generation

Example prompts

  • “/generate-video”

Requirements

  • Python 3

Workflow steps

5 steps, taken from the first numbered list in SKILL.md.

  1. 加载项目和剧本,确认骨架与生成模式一致。
  2. 在分镜图生视频确认分镜图可用;在参考生视频确认视频单元正文非空、编排时长合法。
  3. 调用 MCP 工具入队,处理准入拒绝与档位确认。
  4. 展示结果,按用户选择点名重做不满意的分镜或视频单元。
  5. 以工具写回的 generated_assets.video_clip 作为成片归属。

What it can do on your machine

Read from SKILL.md and the folder at commit 08ab3b3. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Generate Video loads about 1.3k tokens when it runs, and up to ~2.3k if it reads all its reference files. Until then it costs about 15 tokens; SKILL.md has 301 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~15
When it runs · the whole SKILL.md, loaded when a task matches
~1.3k
With references · SKILL.md plus every file in references/, read only if the agent opens them
~2.3k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check: warnings

The automated check found patterns that need a careful read before installing.

  • WarningTells the agent its actions are pre-authorized / not to stop for confirmationSKILL.md:67
    unit_id": "E1S01", "version": 2})`,立即生效,无需用户确认、不收费:

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from ArcReel/ArcReel at commit 08ab3b3, republished under its AGPL-3.0 licence (© ArcReel). 301 words, ~1,254 tokens.

Download SKILL.mdSave it as .claude/skills/generate-video/SKILL.md (or your agent's skills folder). This skill also uses 1 other file; get the full folder from GitHub.
name
generate-video
description
为分镜或自包含视频单元生成视频。当用户要求生成或重做视频时使用;支持整集、单项与批量自选。

生成视频

路由

让 MCP 工具读取 project.json,按 generation_mode × content_mode 分派,并校验剧本骨架:

生成模式×创作类型应有骨架分派输出目录
reference_video × narration / drama / advideo_units[]task_type="reference_video" → execute_reference_video_taskreference_videos/{unit_id}.mp4
storyboard × narrationsegments[]task_type="video" → execute_video_taskvideos/scene_{segment_id}.mp4
storyboard × dramascenes[]同上videos/scene_{scene_id}.mp4
storyboard × adshots[]同上videos/scene_{shot_id}.mp4

骨架失配时停止入队,按项目生成模式重生成剧本。参考生视频直接消费自包含 video_units[],跳过分镜图。

参考生视频

把每个 video_units[] 条目视为一次独立生成调用:

  • 从视频单元正文(text)构造统一引用语法 prompt。
  • 参考图执行期从正文的 @[名称] 按首次提及顺序解析,无特殊排序;有资产图用资产图,否则用该资产的全部原图。
  • 让生成预检把视频单元编排时长投影到供应商申请档位。
  • 遇到 needs_replan 或发声归属问题时停止该视频单元,先修复规划内容。
  • 整集生成只复用 generated_assets.video_clip 明确指向的现行成片;同名孤儿文件不代表该视频单元已完成。

让项目配置、剧本模型与视频能力决定比例、时长和参考图上限,不在调用参数中另写一套数值。

工具调用

使用 MCP 工具入队;本 skill 不提供 Python 或 Shell 生成脚本。

操作工具
整集生成(默认操作)mcp__arcreel__generate_videos({"script": "episode_1.json", "target": {"scope": "episode", "episode_id": 1}})
单分镜mcp__arcreel__generate_videos({"script": "episode_1.json", "target": {"scope": "scene", "ids": ["E1S01"]}})
批量自选mcp__arcreel__generate_videos({"script": "episode_1.json", "target": {"scope": "selected", "ids": ["E1S01", "E1S05", "E1S10"]}})
全部待处理mcp__arcreel__generate_videos({"script": "episode_1.json", "target": {"scope": "all"}})

把 target.ids 在分镜图生视频解释为分镜 ID,在参考生视频解释为 unit_id。整集生成的 target.episode_id 是剧本所属那一集的集 ID(与文件名 episode_{集 ID}.json 中的数字相同,取计划 target.episode),不是第几集。

点名重新生成视频单元

在参考生视频传 video_units[].unit_id:

操作工具
重新生成单个视频单元mcp__arcreel__generate_videos({"script": "episode_1.json", "target": {"scope": "scene", "ids": ["E1U2"]}, "force": true})
重新生成多个视频单元mcp__arcreel__generate_videos({"script": "episode_1.json", "target": {"scope": "selected", "ids": ["E1U2", "E1U3"]}, "force": true})

一次调用完成入队并返回 durable batch;按返回的 poll_after_seconds 调用 get_generation_batch,直到 done: true 后再处理结果:

  • 把点名视为强制重做,覆盖已有成片。
  • 已有在途任务时不自动 force 重做;等待并读取其 batch 结果,避免对刚完成的目标再次付费提交。
  • 只生成剧本中点名的自包含视频单元;未命中的 ID 记为 blocked,带 generation_unit_not_found。
  • 调用中断后查询 durable batch;只把未成功的 ID 用 selected、force: false 重发,已完成项归 skipped。
  • 结果按 requested / succeeded / failed / blocked 逐 ID 返回, 结构与问题码见 .claude/references/generation-results.md。
切换 current 版本

视频单元的 current 版本决定预览、剪辑与导出用哪一版。挑更好的版本时调用 mcp__arcreel__select_video_version({"unit_id": "E1S01", "version": 2}),立即生效,无需用户确认、不收费:

  • 首轮审阅后,某单元的另一候选版本更好,改用它。
  • 重新生成后验收,新版本不如旧版,改回旧版本。
  • 版本号不存在时,按返回的 params.available_versions 重选。
视频与旁白

视频请求只看剧本:一律按视频单元的编排时长申请档位,准入、报价与恢复都与项目的旁白交付方式无关, 未配置 TTS 的项目照常生成与恢复视频。旁白配音在剪辑阶段单独生成(generate-narration-audio)。 generate_videos 没有 narration_delivery 参数,带上会被拒绝。

整批准入判定与档位确认

视频整批请求是全有或全无:准入 admitted 时整批入队,blocked 或 confirmation_required 时 一个任务都不入队。Web 与 Agent 走同一套准入与同一套请求选择语义,没有 Agent 专属的宽松通道。

参考生视频按视频单元的引用状态选择生效档位,把编排时长投影到模型支持的申请档位。申请档位不同于编排时长时 预检返回 reference_duration_confirmation_required,逐档位向用户说明涉及的视频单元、编排秒数、申请秒数 与变长/变短;确认后经 confirmed_request_durations(按 unit_id 记档位)让原目标集合仍作为一批重发:

text
mcp__arcreel__generate_videos({"script": "episode_1.json", "target": {"scope": "episode", "episode_id": 1},
                               "confirmed_request_durations": {"E1U1": 8}})

要在提交前先把费用交给用户确认时(如 edit-video 的勾选清单),带 "preview": true 预检: 同一份准入,不入队,返回逐视频单元的预计费用与档位变化。用户确认后正式提交时原样带上预检给出的 confirmed_request_durations,不会再收到档位确认。

被拒时逐视频单元报告 unit_id、problem.code、原因与 problem.action;通过的视频单元带 generation_batch_admission_withheld,其 blocked_unit_ids 指出是被谁挡住的,如实说明这层因果。 不要把整批拆小去先跑通过的那一半——那既绕开全有或全无,也会重复提交已经付过费的视频单元。 能力无法解析时把工具错误作为 blocker,先修复模型能力声明。

结果怎么读、怎么说

task_state(队列任务)、provider_checkpoint(供应商是否已提交)、artifact_status(产物 current / stale / missing / blocked)与 workflow 步骤状态互相独立,分开陈述:「任务成功」不等于 「当前产物有效」。provider_checkpoint.submitted 为真表示供应商侧很可能已计费;任务 interrupted 表示没有供应商裁决,一律按 problem.action 决定;该情形通常交回 wait_for_task(任务可能仍在跑并正常落地),不要自行改成 retry。

stale 产物照常可预览、可导出、可参与成片,服务端会复用、不会自动重生;是否重做由用户明确决定。 不自动删除、覆盖或重生任何已付费产物与历史版本。

工作流程

  1. 加载项目和剧本,确认骨架与生成模式一致。
  2. 在分镜图生视频确认分镜图可用;在参考生视频确认视频单元正文非空、编排时长合法。
  3. 调用 MCP 工具入队,处理准入拒绝与档位确认。
  4. 展示结果,按用户选择点名重做不满意的分镜或视频单元。
  5. 以工具写回的 generated_assets.video_clip 作为成片归属。

Prompt 构建

让 MCP 工具按生成模式构建 Prompt:

  • 分镜图生视频读取 image_prompt、video_prompt 与分镜图。
  • 参考生视频读取视频单元正文(text)与编排时长。
  • 旁白/解说的分镜图生视频不把 novel_text 放入视频 Prompt;旁白由独立音频流程处理。
  • 自动应用音频开关、角色发声归属与负面 Prompt 规则。

生成前检查

按项目生成模式检查:

  • storyboard:每个目标分镜都有可用分镜图,动作与发声内容可执行。
  • 参考生视频:每个目标视频单元有非空正文、合法编排时长、单一发声归属,且未标记 needs_replan。
  • reference:参考图由服务端在执行期从正文 @[名称] 的首次提及顺序解析;未登记的提及只产生警告、不阻断入队,让服务端按 max_reference_images 裁剪。
  • reference:输出路径为 reference_videos/{unit_id}.mp4。

© ArcReel, AGPL-3.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 1 other file (references) in agent_runtime_profile/.claude/skills/generate-video of ArcReel/ArcReel.

  • SKILL.md
  • references/veo_prompts.md

Open the folder on GitHubat commit 08ab3b3

Compare with similar skills

Generate Video next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Generate Video compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Generate Video this skillArcReel/ArcReel5.4k—~1.3kAutomated safety check: WarnAGPL-3.0
Clipmivo VideoBarneyD66/clipmivo-tools142—~945Automated safety check: PassMIT
ComfyUI Local DriverSlavaSexton/ComfyUI-Agent-Kit105—~12kAutomated safety check: PassApache-2.0
Image Edit Workbenchhenjicc/Henji-AI254—~502Automated safety check: PassApache-2.0
Checkffroliva/gflow-cli266—~2.6kAutomated safety check: PassMIT
OrchestrationOrkas-AI/Orkas-VideoStudio498—~3.4kAutomated safety check: PassMIT

Similar skills

  • Clipmivo Video

    BarneyD66/clipmivo-tools

    Create and manage AI video tasks through ClipmivoAI using its MCP server, CLI or REST API.

    142 GitHub stars~945 tokensUpdated 23 days ago
    Media & CreativeAuto-check passed
  • ComfyUI Local Driver

    SlavaSexton/ComfyUI-Agent-Kit

    Drives a local ComfyUI install over its HTTP API to generate and edit images, video and audio, with per-model prompt recipes and workflow guidance.

    105 GitHub stars~12k tokensUpdated 1 mo ago
    Media & CreativeAuto-check passed
  • Image Edit Workbench

    henjicc/Henji-AI

    在痕迹AI修图、调色、抠出或选中主体、移除物体、修补瑕疵、编辑图层与蒙版、导出图片或流转图片产物时使用。视频时间线与成片用 video-edit-workbench;写代码画面用 video-edit-code-creation。

    254 GitHub stars~502 tokensUpdated today
    Media & CreativeAuto-check passed
  • Check

    ffroliva/gflow-cli

    Auto-fix lint and formatting, then report types and tests. An agent skill from ffroliva/gflow-cli.

    266 GitHub stars~2.6k tokensUpdated yesterday
    Media & CreativeAuto-check passed
  • Orchestration

    Orkas-AI/Orkas-VideoStudio

    The master program for producing or editing a video end to end — read this at the START of any video task (after video-router), then follow the gates and the per-line steps.

    498 GitHub stars~3.4k tokensUpdated 18 days ago
    Media & CreativeAuto-check passed
  • Scenario Runway

    scenario-labs/skills

    A skill your agent uses when generating or editing video with Runway models on Scenario via MCP: text-to-video or image-to-video from a first frame with Gen4.5 (cinematic motion, sequenced actions…

    931 GitHub stars~1.2k tokensUpdated yesterday
    Media & CreativeAuto-check passed

More from ArcReel/ArcReel

All 18 skills in this repo
  • PR AI Review Loop

    ArcReel/ArcReel

    PR AI review 收敛。用户或 team-lead 要求启动或继续审查—修复循环时使用;供本地实现者或受委派的 review-looper 调用,不用于 GitHub reviewer 产出审查意见或仅处理单条评论。

    5.4k GitHub stars~667 tokensUpdated yesterday
    Auto-check passed
  • Tech Doc Style Chinese

    ArcReel/ArcReel

    在撰写、改写、校对或审阅中文技术文档、产品文案、界面文案、Markdown 文档、API 说明、操作手册、故障排查或运维文档时使用。采用克制、准确、可扫读的中文技术写作风格;保留原文事实、限制和机器可读内容;按任务读取术语排版、API 状态文案、项目覆盖或受控中文技术写作参考。

    5.4k GitHub starsUsed in 1 repo~845 tokens
    Auto-check passed
  • Translate Docs

    ArcReel/ArcReel

    Translate every dirty ArcReel documentation source into English and refresh the translation lockfile.

    5.4k GitHub stars~1.8k tokensUpdated yesterday
    Auto-check passed
  • Adapt Custom Endpoint

    ArcReel/ArcReel

    当用户要求把采用 JSON 提交后轮询协议的视频或图片供应商接入 ArcReel,或要求编写、验证、测试、保存自定义调用端点定义时使用。

    5.4k GitHub stars~413 tokensUpdated yesterday
    Auto-check passed
  • Afk Team Workflow

    ArcReel/ArcReel

    把一个 Spec 的全部子 issue(或一组显式 issue)组建团队无人值守跑到全部合并或明确暂停. An agent skill from ArcReel/ArcReel.

    5.4k GitHub stars~794 tokensUpdated yesterday
    Auto-check passed
  • Update Docs

    ArcReel/ArcReel

    根据最近的 git 改动,更新面向用户的文档(README 双语、入门教程、部署、剪映导出等)。手动调用. An agent skill from ArcReel/ArcReel.

    5.4k GitHub stars~544 tokensUpdated yesterday
    Auto-check: notes

Questions about Generate Video

What does Generate Video do?

为分镜或自包含视频单元生成视频。当用户要求生成或重做视频时使用;支持整集、单项与批量自选. An agent skill from ArcReel/ArcReel. Generate Video is an agent skill from ArcReel/ArcReel.

When should I use Generate Video?

Generate Video fits situations like: tasks that involve AI video generation.

How do I install Generate Video in Claude Code?

Run `npx skills add ArcReel/ArcReel --skill generate-video -a claude-code`. Or copy the skill folder (agent_runtime_profile/.claude/skills/generate-video in ArcReel/ArcReel) into .claude/skills/generate-video in your project. Claude Code loads it when a task matches its description.

How do I install Generate Video in Codex?

Run `npx skills add ArcReel/ArcReel --skill generate-video -a codex`. Or copy the skill folder (agent_runtime_profile/.claude/skills/generate-video in ArcReel/ArcReel) into .agents/skills/generate-video in your project. Codex loads it when a task matches its description.

Can I use Generate Video in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add ArcReel/ArcReel --skill generate-video -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/generate-video, .gemini/skills/generate-video, .github/skills/generate-video and .opencode/skills/generate-video in your project.

What does Generate Video need to run?

SKILL.md names no scripts, command-line tools or credentials: Generate Video is instructions for the agent only. Our summary lists: Python 3.

Does Generate Video access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Generate Video safe to install?

Our automated static check of SKILL.md flagged 1 warning(s): tells the agent its actions are pre-authorized / not to stop for confirmation. Read the flagged lines before installing; the check is not a guarantee either way.

What licence does Generate Video use?

Generate Video is published under the AGPL-3.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Generate Video use?

About 1.3k tokens (SKILL.md is roughly 5k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 1.1k tokens, read only when the agent opens those files.

What are the alternatives to Generate Video?

Skills that share tags, products or a category with Generate Video: Clipmivo Video (BarneyD66/clipmivo-tools, 142 stars), ComfyUI Local Driver (SlavaSexton/ComfyUI-Agent-Kit, 105 stars), Image Edit Workbench (henjicc/Henji-AI, 254 stars) and Check (ffroliva/gflow-cli, 266 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Generate Video?

ArcReel (a GitHub organization) maintains it in ArcReel/ArcReel, which has 5,399 GitHub stars. The repository holds 18 skills in this directory. The repository was last updated on October 9, 2026.

Source: ArcReel/ArcReel on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.